J. Geod. Sci.
2015; 5:1–8
Lars E. Sjöberg*
The development of physical geodesy during
1984-2014 – A personal review
Abstract: This article is a personal review of the devel-
opment of physical geodesy during 1984-2014. The pe-
riod is characterized by an intensive advance in both data
and theory to meet the growing technical demands in
GPS/GNSS applications and scientific needs in geoscience.
As a result, many parts of the world are now mapped with a
1 cm detailed geoid model, and the global long- to medium-
wavelengths of the gravity field and geoid are homoge-
neously determined to 1 mGal and 1 cm by satellite-only
dedicated satellite gravity missions. The future can expect
to see even higher demands for accuracy and reliability to
satisfy the specifications for a pure geoid model based ver-
tical datum.
Keywords: geoid; physical geodesy; quasigeoid; topo-
graphic bias
DOI 10.1515/jogs-2015-0003
Received November 5, 2014; accepted February 26, 2015
1 Background
This article is based on an invited presentation at the
Nordic Geodetic Commission (NKG) General Assembly in
Gothenburg in September 2014. The time period consid-
ered in the review corresponds to the period during which I
was a full professor at Royal Institute of Technology (KTH)
in Stockholm (and the only one in Sweden). Naturally my
personal view mirrors and is strongly affected by my expe-
riences in the subject of physical geodesy prior to the time
period. Therefore this presentation will start there.
1.1 My Ph.D. training period 1971-1975
I was a Ph.D. student at KTH between September 1971
and May 1975, and during this time there were still “the
three geodetic kings of Sweden”, who in one way or an-
other contributed to my career. They were L. Asplund
(1914-1987; professor at the Swedish Mapping Authority
*Corresponding Author: Lars E. Sjöberg: Royal Institute of Tech-
nology (KTH) Stockholm, Sweden, E-mail: lsjo@kth.se
© 2015 Lars E. Sjöberg, licensee De Gruyter Open.
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivs 3.0 License.
1949-1975 and National Land Survey 1975-1979; adj. prof.
at KTH 1973-1979; “the father of NKG”), who was respon-
sible for one of my M.Sc. courses, A. Bjerhammar (1917-
2011; professor at KTH 1951-1981), my Ph.D. supervisor, and
E. Tengström (1913-1996; professor at Uppsala University
1968-1978), who was the faculty opponent at my disputa-
tion (and later my scientific advisor and colleague as emer-
itus professor at Uppsala University).
During this period, M.S. Molodensky’s new theory in
physical geodesy became known to us and was much de-
bated. In particular, the concepts of the surface gravity
anomaly, normal height and height anomaly/quasigeoid
were most remarkable. (In fact, already in 1970 the normal
height system was adopted as the Swedish height system.)
Bjerhammar (1962) had already presented his version of
determining the quasigeoid from a finite number of surface
gravity anomalies by analytical continuation to an internal
sphere (“the Bjerhammar sphere”), but this was still not
a well-recognized method (and many geodesists are still
reluctant). At this time, the only influential geodesist that
fully recognized this technique was H. Moritz (Heiskanen
and Moritz 1967, Sect. 8-10). Somewhat later, Molodensky’s
new concept of modifying Stokes’ formula was also appre-
ciated.
A great event was in 1972 when T. Krarup (1919-2005;
from 1949 with the Danish Geodetic Institute in Copen-
hagen) gave a lecture on his report on the fundamental
theory of physical geodesy (Krarup 1969) at Uppsala Uni-
versity. In his study, he set out to prove that Bjerhammar’s
method could not be applied as a reliable approximation
technique. However, during the course of his work he be-
came fascinated about the method, developed it further to
the method of least squares collocation (LSC) and, finally,
came up with a theoretical proof of the validity of analyt-
ical continuation in physical geodesy (the Runge-Krarup
theorem). My own Ph.D. thesis (Sjöberg 1975) was con-
cerned with Bjerhammar’s method, collocation and their
similarities. See also Sjöberg (1978).
1.2 The postdoc period 1977-1984
From early 1977 I spent 15 months as a research associate
at the Ohio State University (OSU) with R.H. Rapp as my
scientific advisor. At that time, OSU was considered as a
Brought to you by | Cinvestav
Authenticated
Download Date | 10/25/17 12:06 AM
2 | Lars E. Sjöberg

young geodesist’s Mecca. Some of the tasks on the phys-

ical geodesy research agenda were applications of satel-
2 Data enhancements
lite altimetry, LSC, development of the OSU Earth Gravita-
Although terrestrial gravity data has been gradually mea-
tional Models (EGMs), topographic problems related with
sured for covering and densifying land areas, the advent
geoid/quasigeoid determination by an EGM (e.g., Sjöberg
of satellite altimetry was the most important advance of
1977) and the requirements needed to achieve a 10 cm
terrestrial data coverage, as it allows gravity mapping on
geoid model. In particular, satellite altimetry was a new
the oceans, which was lacking before. In addition, dur-
and hot research topic. As Dick Rapp had a long-time re-
ing the last decade large regions, both on land and at sea,
search project to develop the OSU EGMs to higher and
have been covered by airborne gravity data. All these new
higher resolution and accuracy, at the time of my arrival he
measurements are complemented with much more accu-
was excited to convert the new altimetry data from satel-
rate positioning by GNSS. In particular, the vertical posi-
lite SEASAT-1 to 1x1 degree2 sea surface gravity anomalies
tion was frequently critical in the past.
by testing the inverse Stokes’ formula (which did not work
In parallel, the long-wavelength gravity field has been
well) and two versions of collocation (Rummel et al. 1977).
explored to unprecedented resolution and accuracy, first
During this period, H. Moritz came on his annual visit to
by dynamic satellite geodesy methods using all kinds of
OSU to write a report in the OSU geodesy series. These re-
geodetic satellite measurements, but finally by dedicated
ports were partly a basis for his text book on advanced
satellite gravity missions. Already during the 1980s and
physical geodesy (Moritz 1980).
90s there were competing technologies for such missions,
Both during my OSU visit and after I met C.C. Tschern-
such as NASA’s Geopotential Research Mission (Yionoulis
ing (1944-2014), who asked why I had started working on
and Pisacane 1985), based on the satellite-to-satellite
geoid estimation by modifying Stokes’ formula instead of
tracking technique (e.g., Sjöberg 1982) and ESA’s ARIS-
exploring LSC. I declared my opinion that the modification
TOTELES mission using satellite gradiometry. It was not
of Stokes’ formula is more far reaching than LSC. In par-
before the millennium shift that such gravity measure-
ticular, I could see computational problems for large data
ment tools were realized (through CHAMP, GRACE and
sets as well as problems with the covariance representa-
GOCE satellite missions). In this way, the long-wavelength
tions. LSC is a fantastic tool for interpolation and to ap-
gravity field has been recovered to a resolution of about
proximate solutions, but I felt there were problems in ob-
100 × 100 km2 , providing gravity and geoid accuracies of
taining the most accurate solutions. This discussion and
the orders of 1 mGal and 1 cm, respectively.
comparison of the two methods continued now and then
In the meantime satellite data was combined with ter-
through the 1980s.
restrial gravity data to high-resolution EGMs. The most
The satellite geodesy symposium in Cannes in Decem-
recent one is EGM2008 (Pavlis et al. 2012), complete
ber 1980 was a great event for me, as it was the first time
to degree and order 2159, corresponding to a resolution
I gave an international talk on least squares combination
of the order of 5’x5’. In parallel, high-resolution Digi-
(limited to simple spectral combination) of satellite and
tal Elevation Models have also been developed based on
terrestrial gravity data for geoid determination (Sjöberg
new terrestrial data, as well as space data. Examples are
1981). After my presentation, I had a long discussion with
the height databases DTM2006 (Pavlis et al. 2006) and
H-G Wenzel, who explored spectral combination further
ETOPO1 (Amante and Eakins 2009) to resolutions of 5’×5’
(Wenzel 1981 and 1982).
and 1’×1’, respectively.
During spring 1983, I visited Stuttgart University as an
Alexander von Humboldt fellow with E Grafarend as my
host. My research task was to study variance component
estimation, in particular non-negative variance estima- 3 Development of theory
tion. Together with my previous assistantship for Bjerham-
mar on his research on the theory of errors (Bjerhammar 3.1 The general principles for geoid
1973), this study has been important for my future interest determination
in applying least squares, e.g., in physical geodesy.
During the 1980s, the goal for the accuracy in geoid mod-
elling was of the order of 10 cm. This was to some extent
beyond the goal of the textbook of Moritz (1980), in par-
ticular as GPS applications usually required even higher
accuracy. However, there was and still is no doubt that the

Brought to you by | Cinvestav

Authenticated
Download Date | 10/25/17 12:06 AM
 The development of physical geodesy during 1984-2014 – A personal review
high precision geoid models can only be achieved by com-
bining the well determined long-wavelength gravity field,
determined by an EGM with the details determined by a
dense terrestrial gravity anomaly grid. A number of such
models were developed. An early candidate was by LSC.
However, during the 80s various types of modified Stokes’
formulas were also developed, originating with M. S. Molo-
densky’s method (Molodensky et al. 1962, Ch. 7). These
formulas mainly differ by the deterministic choice of the
modification parameters in Stokes’ modified kernel func-
tion, and by the cap size/area of truncating Stokes’ inte-
gral. The main principle is to choose the parameters such
that the truncation error is as small as possible. However,
among these methods only Molodensky’s method uses a
formal principle of optimization to achieve this goal. An
alternative strategy is to select the modification parame-
ters such that the expected geoid model error due to trun-
cation error and erroneous terrestrial and satellite gravity
data is as small as possible by applying a least squares
technique. The original method of spectral combination
of Sjöberg (1981) and Wenzel (1981) and (1982) assumes
that there is no truncation of the integration area, and the
solution therefore is a simple spectral weighting of satel-
lite and terrestrial gravity data w.r.t. their a priori stan-
dard errors. The methods presented by Sjöberg (1984a,
b) and (1991) also consider the truncation error, and this
technique was further developed to the Least Squares
Modification of Stokes formula with Additive corrections
(LSMSA) by Sjöberg (2003a, b).
A different way of modifying Stokes’ formula uses a
higher order reference field, determined by an EGM (e.g.,
Vanicek and Sjöberg 1991). That is, only a residual grav-
ity anomaly (with the EGM contribution removed) is used
in Stokes’ formula, and the EGM contribution is restored
directly as a long-wavelength geoid undulation. It can be
shown (Sjöberg 2005, Appendix) that theoretically and for
the same modification parameters (disregarding practical
issues) there is no difference between this technique and
the modification described above, e.g., in Sjöberg (2003a).
A special version of this technique is called Remove-
Compute-Restore (RCR), implying that not only the EGM
part of the gravity anomaly is removed under Stokes’ inte-
gral, but also much of the high frequency signal as a DEM
representation of the attraction of gravity (e.g., Forsberg
1993; Sideris and Forsberg 1990). Usually this method does
not apply a modification of Stokes’ kernel, which makes
the method prone to truncation bias (Sjöberg 2005). This
problem can be avoided by using a sufficiently large inte-
gration region at the prize of possible increase in system-
atic long- to medium-wavelength errors propagated from
the terrestrial data. Typically various topographic, atmo-
| 3
spheric and ellipsoidal corrections are rather crudely ap-
proximated, which may also bias the solutions (see e.g.
Sjöberg 1999, 2000 and 2003c). For example, the Stokes
types methods presented in Sansó and Sideris (2013) seem
to follow the traditional line of thought, lacking the re-
finements needed in today’s “one-cm geoid”, at least in
mountainous regions. (See also Sects. 3.2 and 4.3.) How-
ever, the UNB way of applying a higher order reference
field together with a Molodensky type of modification of
Stokes’ formula is used with rigorous corrections (Ellmann
and Vanicek 2007; Vanicek et al. 2013).
3.2 Trends for geoid corrections
Since the 1980s, most geodesists apply either LSC or one
form of the RCR technique (frequently with FFT to speed
up the computations), probably influenced by the IAG
geoid schools and its lecture notes propagated also into
the textbook of Sansó and Sideris (2013). In the RCR meth-
ods, the observed gravity is usually, following the tradi-
tional technique, being corrected for the topographic at-
traction and downward continued to the geoid by a sim-
ple free-air correction to create the gravity anomaly on the
geoid by removing normal gravity on the reference ellip-
soid (e.g., Tziavos and Sideris 2013). By further adding a
topographic compensation attraction and the secondary
indirect topographic effect (SITE), the topographically re-
duced gravity anomaly is ready for Stokes’ integration, and
the product is the co-geoid height, which needs a correc-
tion for the Primary Indirect Topographic Effect (PITE) to
provide the final geoid height. Sjöberg (2014) and (2015)
showed that this technique suffers from the rough down-
ward continuation by the free-air correction, which yields
an error of the order of the geoid-to-quasigeoid difference
(GQD), which may range to a few metres.
Alternatively, the modern Molodensky type of surface
gravity anomaly is applied. In Vanicek and Kleusberg
(1987) the downward continuation to sea level was sim-
ply neglected. Sideris and Forsberg (1990) and Sansó and
Sideris (2013) provided a method for a Helmert type of
compensation for the topographic reduction, supported
by Wang and Rapp (1990) but debated by Martinec et al.
(1993). Sjöberg (2015) concludes that there are two major
problems with this technique: a) the free-air correction is
at least as inaccurate now as in the classical approach
above, and b) the surface gravity anomaly after applying
the Direct Topographic Effect on gravity (DITE) is not a rig-
orous gravity anomaly suitable for dwc (see Sect. 4.3). As
a result, this gravity anomaly yields a geoid estimator at
Brought to you by | Cinvestav
Authenticated
Download Date | 10/25/17 12:06 AM
4 | Lars E. Sjöberg
least as bad as the one provided by the classical gravity
anomaly above.
A correct instruction for the topographic correction of
the surface gravity anomaly and its use for geoid deter-
mination was presented already in Heiskanen and Moritz
(1967, Sect. 8-11). The DITE of the surface gravity anomaly
and the SITE are applied at the Earth’s surface, a) the re-
duced gravity anomaly is rigorously downward continued
to the geoid level (e.g. by solving Poisson’s integral equa-
tion), and, b) after Stokes’ integration, the PITE is applied.
This instruction is followed in the UNB and LSMSA tech-
niques.
For accurate geoid determination, a small correction
must also be applied for the atmospheric effect caused
by the forbidden atmospheric masses in Stokes’ integra-
tion. This is traditionally performed by the IAG technique
(Moritz 1984) by applying a simple correction to the ob-
served gravity anomaly prior to Stokes’ integration, and
possibly adding a small indirect atmospheric effect to the
result. This procedure works well for global integration,
but when using a limited area of integration (which is
always the case in practice) a bias occurs which grows
with the integration area to a maximum of more than 3 m
(Sjöberg 1999). As shown by Sjöberg (ibid), the bias stems
from the ignored zero-degree normal gravity anomaly (of
the order of 0.87 mGal), which contributes more or less
to the truncated Stokes integral dependent on the limit of
the integration area. This problem is easily solved in the
LSMSA method, where all geoid corrections are added and
combined with direct and indirect effects to the prelimi-
nary geoid heights (Sjöberg 2003b). However, the atmo-
spheric effect is usually negligible when using a satellite-
only EGM and a small cap size in Stokes integration. This
holds also for the ellipsoidal effect (e.g., Ellmann and
Sjöberg 2004 and Sjöberg 2004b).
4 Other theoretical developments
The development of the following four theoretical issues
in physical geodesy will be reviewed in this section: the
determination of the potential of the geoid (W0), the topo-
graphic bias in analytical continuation, the GQD and the
SITE.
4.1 The potential of the geoid
The NKG geoid models are regional geoid models, imply-
ing that W0 is of less importance than for absolute geoid
determination. In any case, W0 is closely related with
the zero-degree term of the geoid height, which is a kind
of scaling parameter for the geoid model which needs a
known distance for its calibration. Today W0 is primar-
ily determined from satellite altimetry on the oceans in
combination with an EGM (e.g., Dayoub et al. 2012). There
are also strategies to use the EGM globally, either in di-
rect estimates of W0, or indirectly after first determining
the parameters of the reference/globally best fitting ellip-
soid. Both these global methods suffer from the common
problem that they rely on a prior estimated absolute geoid
model (which thus includes the unknown W0). Sjöberg
(2013b) proposed a least squares approach to determine
the geometric parameters of the reference ellipsoid and
W0 simultaneously.
4.2 The topographic bias
Bjerhammar’s (1962) approach is to analytically continue
the surface gravity anomalies to an inner sphere (the
Bjerhammar sphere), where the traditional formulas of
Poisson, Stokes and Vening Meinesz are applied to de-
termine the gravity anomaly, height anomaly and deflec-
tions of the vertical at any point on or outside the sur-
face of the Earth. The harmonic downward continuation
of the gravity anomaly into the masses makes the grav-
ity anomaly on the Bjerhammar sphere fictitious, implying
that Stokes’ formula applied with these anomalies on the
inner sphere yields biased disturbing potentials and geoid
heights (while the quantities determined on or outside the
Earth’s surface by the extended formulas are unbiased).
Sjöberg (2007a) and (2009a-c) proved that for a constant
topographic density (ρ), the bias in the geoid height equals
2H 3
(︂ )︂
T 2πGρ 2
dN = H + , (1)
𝛾0 3R
where H is the orthometric height, G is the gravitational
constant, 𝛾0 is normal gravity on the reference ellipsoid
and R is the mean sea level radius. As Eq. 1 is the geoid con-
tribution from a Bouguer shell, it leads to the fact that the
terrain correction is not needed on geoid determination
(Sjöberg 2009b), but the topographic correction is purely
dependent on the topographic density distribution along
the vertical at the computation point. This technique is
employed on the LSMSA method for geoid determination
(Sjöberg 2003b). Eq. 1 also allows a simple determination
of the geoid change for a laterally changing topographic
density (Sjöberg 2004).
Brought to you by | Cinvestav
Authenticated
Download Date | 10/25/17 12:06 AM
 The development of physical geodesy during 1984-2014 – A personal review
4.3 The DITE and SITE
As discussed in Sect. 3.1 the reduction of the surface grav-
ity anomaly for topography requires that the topographic
attraction as well as the SITE be applied at surface level.
The combined correction is the DITE of the surface gravity
anomaly, while the correction for the topographic attrac-
tion alone is the DITE of gravity and gravity disturbance.
This can be understood from the fact that the gravity dis-
turbance is an attraction (i.e. the gradient of a potential),
while the gravity anomaly is not (Vajda et al. 2006; Sjöberg
2013c and 2014, 2015). Only the gravity anomaly reduced
for both these effects is a rigorous gravity anomaly with all
topographic signals removed, suitable for dwc in free air
without bias. Possible topographic compensation should
also be applied at the surface point, both for attraction and
SITE. As a result, the traditional definition of the SITE to
provide a correction of the gravity anomaly from the geoid
to the co-geoid (Heiskanen and Moritz 1967, p. 142) should
be replaced by the new definition that it is a part of the
DITE of any gravity anomaly in space. Only after this cor-
rection, the gravity anomaly can be upward or downward
continued in free space above sea level.
4.4 The GQD
Most Nordic countries use normal heights for their verti-
cal height systems, implying that the quasigeoid is the pri-
mary geoid type to consider. Nevertheless, the real geoid
is always of interest as well, not the least for geophysical
applications. In the last few years, the traditional formula
for converting the quasigeoid height to the geoid height by
the Bouguer gravity anomaly (times orthometric height di-
vided by mean normal gravity at the computational point;
Heiskanen and Moritz 1967, Sect. 8-13) has been consid-
erably improved. First by a term that considers the topo-
graphic potential difference at the geoid and surface point
(Flury and Rummel 2009), and then by an additional cor-
rection term that is assumed to be small (Sjöberg 2010). Fi-
nally, Sjöberg (2012) showed the GQD for an arbitrary com-
pensation of the topographic attraction and potential.
5 NKG geoid models
The NKG working group on geoid determination with
members from each Nordic country except Iceland was
established in 1978. Most of the time geodesists from the
Baltic States contributed to the work. Originally, the group
embarked on the method of LSC with the aim of determin-
| 5
ing the geoid to an accuracy of 0.5 m (Tscherning 1983).
However, after completing the first official NKG model
(Tscherning and Forsberg 1986), the use of collocation
was limited to interpolation of gravity anomalies, and
the much more efficient FFT technique was employed for
the geoid computation. This model was succeeded by im-
proved models in 1989 (Forsberg 1990) and 1996 (Forsberg
1999), which used the FFT and RCR techniques. In 1989,
the expected accuracy was of the order of 10 cm. Since 1999
the group was struggling towards “the 1-cm geoid”, pri-
marily by filling-in various regions with gravity data gaps.
These efforts did not lead to substantially improved mod-
elling of the geoid (Forsberg 2001), whereupon I suggested
in 2002 that also the theory might need some improve-
ments. In the next model presented to the NKG General
Assembly in 2004, Stokes’ kernel was finally modified (ac-
cording to Wong and Gore 1969). For further details, see
Sjöberg and Ågren (2002) and Sjöberg (2005).
Currently the NKG geoid working group is in the stage of
testing and comparing various computational techniques,
such as the LSMSA method and different versions of RCR.
[Recent comparisons of the RCR and LSMSA techniques
can be found, e.g., in Ågren et al. (2009) and Yildiz et al.
(2012).] A new NKG geoid model at the expected accuracy
of 1 cm for major parts of the region is planned to be pre-
sented in 2015. In parallel with this task, a project to in-
vestigate the requirements for a 5 mm geoid model in the
Nordic countries is under way (Ågren and Sjöberg 2013 and
2015).
6 Gravity inversion
Gravity inversion is important both in Geodesy and Geo-
physics. After the advent of satellite gravimetry, various
applications necessitate the dwc of the space data to or be-
low the Earth’s surface, and such solutions are smoothed
when using an RCR technique for the topographic effect.
If gravity is needed for gravity inversion at sea level, e.g.
for geoid determination over land areas, for geophysical
studies of Moho geometry and density contrast and/or for
other density structures below the Earth’s crust, the grav-
ity signal of the topography should be removed from the
data. Traditionally, it is assumed both in Geodesy and Geo-
physics that the (refined) Bouguer gravity anomaly fulfils
this requirement. However, as pointed out by Vajda et al.
(2006), Sjöberg (2013c) and Sjöberg et al. (2014) this is not
the case, but the SITE must also be applied to achieve the
rigorous no-topography (NT) gravity anomaly. Note that
Brought to you by | Cinvestav
Authenticated
Download Date | 10/25/17 12:06 AM
6 | Lars E. Sjöberg
also the Bouguer gravity disturbance is a no-topography
quantity.
In Moritz (1990, Sect. 8.3.2), Vening Meinesz’ traditional
isostatic hypothesis based on the condition that the iso-
static gravity anomaly vanishes for the isostatic equilib-
rium in a regional isostatic compensation was generalized
to a global compensation mechanism. In addition, the flat
Earth approximation used in Vening Meinesz’ model was
refined to a spherical Earth model. The Vening Meinesz-
Moritz (VMM) isostatic problem to determine the Moho
depth was presented by Sjöberg (2009d) as that of solv-
ing a non-linear Fredholm integral equation of the first
kind. Sjöberg also presented a first-order solution based
on the Bouguer gravity anomaly, preferably represented
in terms of an EGM. Sjöberg and Bagherbandi (2011) al-
tered the solution to that for the Moho density contrast,
as well as for solving both the Moho depth and density
contrast in a least squares procedure from preliminary de-
termined gravimetric and seismic estimates. The theory
has been applied in several studies, such as Sjöberg and
Bagherbandi (2011) and Bagherbandi and Sjöberg (2013)
to mention a few. Typically, the Moho depth models based
on gravimetric and seismic data agree to the order of a few
kilometres, and even better when reducing the gravity sig-
nal for non-isostatic effects (e.g. Bagherbandi and Sjöberg
2013).
Tenzer and Bagherbandi (2012) compared the global
Moho geometry models based on Bouguer gravity anoma-
lies and disturbances, and they could show that the result
from gravity disturbances agreed significantly better with
a seismic model. Sjöberg (2013) and Sjöberg et al. (2014)
presented theoretical proofs, and they also showed numer-
ically that the Bouguer gravity disturbance provides re-
sults equivalent to those from the NT gravity anomaly. As
the rigorous gravity anomaly (e.g., the NT anomaly) needs
the SITE correction, the rigorous isostatic gravity anomaly
needs a Secondary Indirect Isostatic Effect (SIIE).
7 Concluding remarks
The time interval 1984-2014 constitutes a very exciting pe-
riod of development in the history of Physical Geodesy.
This fast development was encouraged by the need for
accurate geoid models in the new technology GPS sur-
veying, but the progress was also pushed forward by sci-
entific needs, in particular in geophysics (oceanography)
and climate change. Thanks to science’s needs, three suc-
cessful dedicated satellite gravity missions could be re-
alized in the first decade of the new millennium. How-
ever, besides new data, there was also a remarkable devel-
opment in technology and applications for gravity map-
ping on land, air and at sea. For instance, we have seen
a fast development of EGMs, suitable for global and re-
gional geoid modelling. In particular EGM 2008, complete
to degree and order 2159, corresponding to a resolution of
the order of 9 km on the Earth’s surface, should be men-
tioned. The combination of an EGM with terrestrial gravity
data allows the determination of a detailed geoid model
to about 1 cm, and on-going work attempts to specify the
requirements for the 5 mm geoid, which will definitely be
needed once nations give up vertical benchmarks to be
replaced by a geoid model surface as the basis for their
height systems. I believe that the LSMSA and possibly the
UNB geoid modelling techniques are candidates, but the
high accuracy goal cannot be met just by data develop-
ment, but also theory must be further improved. For exam-
ple, as the UNB method is an RCR technique, the possible
atmospheric bias must be considered. The LSMSA method
is unique and has special advantages as it a) optimizes the
expected mean square error of the geoid height w.r.t. trun-
cation bias and errors in gravity and EGM data, b) provides
a least squares estimate for the expected mean square er-
ror, c) has numerical advantages in computing the additive
corrections for topography, atmosphere and ellipsoidal ef-
fects, d) allows simple presentations of each of the addi-
tive effects and e) is flexible in view of that the geoid solu-
tion can easily be improved if some data for additive cor-
rections is improved. For example, the very simple formula
to determine the combined topographic effect without the
need for terrain correction is notable.
Among the remaining problems to be solved is the bias
in gravity field data caused by the Sea Surface Topography
(SST) signal in satellite altimetry. As there is no successful
independent way to remove the SST of the order of ±1−2 m
from the altimetry data, this bias remains more or less in
the products.
Another question to be sorted out by nations is whether
the normal height or orthometric height is the best ba-
sis for their future height systems. There is no doubt that
the normal height and the related height anomaly are bet-
ter realized than the orthometric height and the geoid,
which depend on uncertain topographic density distribu-
tion (Sjöberg 2013a), while others like Vanicek et al. (2012)
are of opposite opinion. Admittedly, the geoid is frequently
a much smoother surface, suitable for interpolation and
detailed presentation than the quasigeoid, which is hard
to visualize in rough terrain with sometimes even a mul-
tiple or singular location for a specific latitude and lon-
gitude. Nevertheless, the uncertainty in the topographic
mass distribution is likely to make the geoid and orthomet-
Brought to you by | Cinvestav
Authenticated
Download Date | 10/25/17 12:06 AM
 The development of physical geodesy during 1984-2014 – A personal review
ric height less attractive versus the quasigeoid and normal
height.
A notable problem is to provide funding for students
and specialists in physical geodesy. Accurate geoid mod-
els are required, e.g. by all GNSS users, who want accurate
3-D positions, but few users or organizations are ready to
pay for them. It is obviously not too difficult to convince
money holders to contribute to new data collections, while
getting funds for personnel to get skilled in enhancing the-
ory is much more of a headache. This problem is probably
due to the fact that physical geodesy is a small and theo-
retically hard-to-grasp discipline, thus there are not many
of its “friends” in funding organizations. As with the rest
of the international geodetic community (IAG), physical
geodesy has problems to motivate its existence, and many
of its important products are taken for granted in science
and society, not the least among GNSS users.
References
Amante C, Eakins B W, 2009, ETOPO1 Arc-Minute Global Relief Model:
Procedures, Data Sources and Analysis. NOAA Technical Memo-
randum NESDIS NGDC-24
Ågren J, Sjöberg L E, 2013, Errors in geoid and quasigeoid models as
propagated from systematic uncertainties in the digital elevation
model. Presented at the IAG Conference, Potsdam, October 1-6,
2013
Ågren J, Sjöberg L E, 2014, Investigations of the requirements for a fu-
ture 5 mm quasigeoid model over Sweden. In U Marti (Ed.): Gravity,
geoid and height systems, Springer Symposia, pp. 143-150.
Ågren J, Sjöberg L E and Kiamehr R, 2009, The new gravimetric geoid
model KTH08 over Sweden. J of Applied Geodesy 3: 143-153
Bagherbandi M, Sjöberg L E, 2013, Improving Gravimetric-Isostatic
Models of Crustal Depth by Correcting for Non-Isostatic Effects and
Using CRUST2.0, Earth-Science Reviews 117(2013): 29-39
Bjerhammar A, 1962, Gravity reduction to an internal sphere. Royal
Institute of Technology, Division of Geodesy, Stockholm, Sweden
Bjerhammar A, 1973, Theory of errors and generalized matrix in-
verses. Elsevier Scientific Publ. Co.
Dayoub N, Edwards S J, More P, 2012, The Gauss-Listing potential
value W0 and its rate from altimetric mean sea level and GRACE.
J Geod 86: 681-694
Ellmann A, Sjöberg LE, 2004, Ellipsoidal correction for the modified
Stokes formula. Boll Geod Sci Aff 63: 153-172
Ellmann A, Vanicek P, 2007, UNB application of Stokes-Helmert‘s ap-
proach to geoid computation, J Geodyn 43: 200-2013
Flury J, Rummel R, 2009, On the geoid-quasigeoid separation in
mountain areas. J Geod 83: 829-847
Forsberg R, 1990, A new high-resolution geoid of the Nordic area. In
Rapp R H, Sanso F (Eds.): Determination of the geoid. IAG Sympo-
sium 106, Springer
Forsberg R, 1993, Modelling the fine structure of the geoid: methods,
data requirements and some result. Surveys in Geophys 14: 403-
418
| 7
Forsberg R, 1999, The NKG-1996 geoid. In Jonsson B (Ed.): Proc. 13th
general meeting of the NKG, The National Land Survey, Gävle, Swe-
den
Forsberg R, 2001, Development of a Nordic cm-geoid – with basics
of geoid determination. In Harsson B G (Ed.): Lecture notes for
NKG autumn school, 28 Aug.-2 Sept., 2000, Fevik, Norway, Statens
Kartverk, Hönefoss, Norway
Heiskanen WA, Moritz H, 1967, Physical Geodesy, W H Freeman and
Co, San Francisco and London
Krarup T, 1969, A contribution to the mathematical foundation of
physical geodesy. Geodaetsk Insititut, Meddelelse No. 44, Copen-
hagen. Also in K Borre (Ed.): Mathematical Foundation of Geodesy-
Selected papers of T Krarup, Springer, 2006.
Martinec Z, Matyska C, Grafarend E W, Vanicek P, 1993, On Helmert’s
2nd condensation method. Manuscr Geod 18: 417-421
Molodensky MS, Eremeev VF, Yurkina MI, 1962, Methods for Study
of the External Gravitational Field and Figure of the Earth. Tranls.
From Russian (1960), Israel program for Scientific Translations,
Jerusalem, Israel
Moritz H, 1980, Advanced Physical geodesy. Herbert Wichmann Ver-
lag, Karlsruhe
Moritz H, 1984, The geodetic reference system 1980. Bull Geod 58(3):
388-398
Moritz H, 1990, The figure of the earth. Wichmann, Karlsruhe
Pavlis N K., Factor K, and Simon A. Holme S A, 2006, Terrain-Related
Gravimetric Quantities Computed for the Next EGM. Presented at
the 1st International symposium of the International gravity ser-
vice 2006, August 28-September 1, Istanbul, Turkey.
Pavlis N A, Simon A H, Kenyon S C, Factor J K, 2012, The Development
and Evaluation of the Earth Gravitational Model 2008 (EGM2008).
JGR 117, B04406
Rummel R, Sjöberg L, Rapp R H, 1977, The determination of gravity
anomalies from geoid heights. Dept Geod Sci Report No. 269, OSU,
Columbus, OH
Sansó F, Sideris, M G (Eds.), 2013, Geoid determination - Theory and
Methods. Springer
Sideris M, Forsberg R, 1990, Review of prediction methods in moun-
tainous regions. In Rapp, R H, Sanso F (Eds.): Determination of the
geoid, present and future, IAG Symposia 106, Springer.
Sjöberg LE, 1975, On the Discrete Boundary Value Problem of Physi-
cal Geodesy with harmonic Reduction to an Internal Sphere. Ph.D.
Dissertation, KTH, Stockholm
Sjöberg L E, 1977, On the Errors of Spherical Harmonic Developments
of Gravity at the Surface of the Earth. Department of Geodetic Sci-
ence, Report No 257, OSU, Columbus, OH.
Sjöberg LE, 1978, A Comparison of Bjerhammar’s Methods and Collo-
cation in Physical Geodesy. Dept Geodetic Science Report No. 273,
OSU, Columbus, OH
Sjöberg L E, 1981, Least Squares Combination of Satellite and Terres-
trial Data in Physical Geodesy. Ann. Geophys., 37(1): 25-30
Sjöberg L E, 1982, On the Recovery of Geopotential Coeflcients us-
ing Satellite-to-Satellite Range-Rate Data on a Sphere. Bull Geod
56(1): 27-39
Sjöberg LE, 1984a, Least squares modification of Stokes’ and Ven-
ing Meinesz’ formulas by accounting for truncation and potential
coeflcient errors. manuscr geod 9: 209-229
Sjöberg LE, 1984b, Least squares modification of Stokes’ and Vening
Meinesz’ formulas by accounting for errors of truncation, potential
coeflcients and gravity data. Dept of Geodesy Rep No 27, Univ of
Uppsala, Uppsala, Sweden
Brought to you by | Cinvestav
Authenticated
Download Date | 10/25/17 12:06 AM
8 | Lars E. Sjöberg
Sjöberg LE, 1991, Refined least squares modification of Stokes’ for-
mula. manuscr geod 16: 367-375
Sjöberg LE, 1999, The IAG approach to the atmospheric geoid correc-
tion in Stokes formula and a new strategy, J. of Geodesy, 73(7):362-
366
Sjöberg LE, 2000, On the topographic effects by the Stokes-Helmert
method of geoid and quasi-geoid determinations. J Geod 74(2):
255-268
Sjöberg L E, 2003a, A general model for modifying Stokes’ formula
and its least-squares solution. J Geod 77: 459-464
Sjöberg LE, 2003b, A computational scheme to model the geoid by
the modified Stokes’s formula without gravity reductions. J Geod
77(2003): 423-432
Sjöberg LE, 2003c, A solution to the downward continuation effect on
the geoid determined by Stokes’ formula. J. Geod 77: 94-100
Sjöberg LE, 2003d, The ellipsoidal corrections to order e2 of geopo-
tential coeflcients and Stokes’ formula. J Geod 77: 139-147
Sjöberg LE, 2004, The effect on the geoid of lateral topographic den-
sity variations. J Geod 78:34-39
Sjöberg LE, 2005, Discussion on the approximations made in the
practical implementation of the remove-compute-restore tech-
nique in regional geoid modelling. J Geod 78: 645-653
Sjöberg LE, 2007, The topographic bias by analytical continuation in
physical geodesy. J Geod 81: 345-350
Sjöberg LE, 2009a, The terrain correction in gravimetric geoid deter-
mination – is it needed? Geophys J Int 176:14-18
Sjöberg LE, 2009b, On the topographic bias in geoid determination
by the external gravity field, J Geod 83:967-972
Sjöberg LE, 2009c, Solving the topographic bias as an Initial Value
Problem. Art. Sat. 44 (3):77-84
Sjöberg LE, 2009d, Solving Vening-Meinesz-Moritz inverse problem
in isostasy, Geophys. J Int. 179: 1527-1536
Sjöberg L E, 2010, A strict formula for geoid-to-quasigeoid separa-
tion. J Geod 84: 699-702
Sjöberg L E, 2011a, Quality estimates in geoid computation by
EGM08. JGS 1(2011): 361-366
Sjöberg L E, 2011b, Local least squares spectral filtering and combina-
tion by harmonic functions on the sphere. J Geod Sci 1(2011):355-
360
Sjöberg L E, 2012, The geoid-to-quasigeoid difference using an arbi-
trary gravity reduction model. Stud Geophys Geod 56: 929-933
Sjöberg LE, 2013a, The geoid or quasigeoid- which reference surface
should be preferred for a national height system? J Geod Sci 3:
103-109
Sjöberg L E, 2013b, New solutions for the geoid potential W0 and the
Mean Earth Ellipsoid dimensions. J Geod Sci 3(2013): 258-265
Sjöberg, L E, 2013c, On the isostatic gravity anomaly and disturbance
and their applications to Vening Meinesz-Moritz inverse problem
of isostasy, Geophys J Int 193: 1277-1282
Sjöberg L E, 2014, On the topographic effects by Stokes’ formula. J
Geod Sci 4: 130-135
Sjöberg L E, 2015, The secondary indirect topographic effect in phys-
ical geodesy, Stud Geophys Geod (on-line)
Sjöberg L E, Bagherbandi M, Tenzer R, 2015, On gravity inversion by
the no-topography and rigorous isostatic gravity anomalies, Pure
and Appl. Geophys. (on-line)
Sjöberg L E, Ågren J, 2002, Some problems in the theory used for the
NKG geoid model. In M. Poutanen and H. Suurmäki (Eds.): Proc.
14th General Meeting of the Nordic Geodetic Commission, Publ.
from the Finnish Geodetic Institute, 2002, pp. 121-130.
Sjöberg L E, Bagherbandi M, 2011, A method of Estimating the
Moho density contrast with a tentative application of EGM08 and
CRUST2.0. Acta Geophys (2011)59(3): 502-525
Tenzer R, Bagherbandi M, 2012, Reformulation of the Vening
Meinesz-Moritz inverse problem of isostasy for isostatic gravity
disturbances. Int J Geosci 3: 918-929
Tscherning C C, 1983, Determination of a (quasi) geoid for the Nordic
countries from heterogeneous data using collocation. In: Proc. 9th
General Meeting of the NKG, Sept 1982, Swedish National Land
Survey, Gävle, Sweden.
Tscherning C C, Forsberg R, 1986, Geoid determination in the Nordic
countries- a staus report. In Kakkuri J (Ed.): Proc. 10th general
Meeting of the NKG, Helsinki, 29 Sept.- 3 Oct., 1986, Finnish
Geodetic Institute, Helsinki
Tziavos I N, Sideris M, 2013, Topographic reductions in gravity and
geoid modeling, In: Sansó and Sideris, M (Eds.) Geoid determi-
nation - Theory and Methods, Lecture Notes in Earth System Sci-
ences, Springer, Ch. 8.
Vajda P, Vaníček P, Meurers B, 2006, A new physical foundation for
anomalous gravity. Stud Geophys Geod 50(2): 182-216
Vanicek P, Kingdon R and Santos, 2012, Geoid versus quasigeoid: a
case of physics versus geometry, Contr. Geo- phys. Geod. 42, 1,
101-117.
aníček P, Kleusberg A, 1987, The Canadian geoid – Stokesian ap-
proach. manusc geod 12: 86-98
Vaníček P, Sjöberg L, 1991, Reformulation of Stokes’s theory for
higher than second-degree reference field and modification of in-
tegration kernels, J Geophys Res 96(B4): 6529-6540
Vaníček P et al., 2013, Testing Stokes-helmert geoid model compu-
tation on a synthetic gravity field: experiences and shortcomings.
Stud Geophys Geod 57: 369-400
Wang Y M, Rapp R H, 1990, Terrain effects on geiod undulation com-
putations. Manuscr Geod 15: 23-29
Wenzel H-G, 1981, Zur Geoid bestimmung durch Kombination von
Schwereanomalien und einem Kugelfunktionsmodell mit hilfe von
Integralformeln. ZfV 106(3): 102- 111
Wenzel H-G, 1982, Geoid computation by least squares spectral com-
bination using integral kernels. Presented paper, IAG General Met-
ing, Tokyo.
Wong L, Gore R, 1969, Accuracy of geoid heights from modified
Stokes kernels. Geophys J R astr Soc 18: 81-91
Yildiz H, Forsberg R, Ågren J, Tscherning CC, Sjöberg LE, 2012, Com-
parison of remove-compute-restore and least squares modifica-
tion of Stokes’ formula techniques to quasi-geoid determination
over the Auvergne test area. J Geod Sci 2(2012): 53-64.
Yionoulis, S M, Pisacane, V L, 1985, Geopotential research mission:
status report. IEEE Transactions on Geoscience and Remote sens-
ing, Vol. GE-23 (4)
Brought to you by | Cinvestav
Authenticated
Download Date | 10/25/17 12:06 AM